By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
443,665 Members | 1,251 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 443,665 IT Pros & Developers. It's quick & easy.

Peoplesoft on Federated UDB?

P: n/a
Anyone using Peoplesoft on a Federated UDB
(shared nothing)Environment on Open System Platforms?
Preferably AIX, but any war stories would be good.

TEA
EB-C
Nov 12 '05 #1
Share this Question
Share on Google+
96 Replies


P: n/a
Bad Pony,

Do you mean federated as in using Nicknames to access other DBMS (DB2 or
non-DB2) or do you mean DB2 ESE with DPF (shared nothing paralelism with
a single-system image, i.e. one DB2 database with partitioned tables)?

Cheers
Serge
Nov 12 '05 #2

P: n/a
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
Bad Pony,

Do you mean federated as in using Nicknames to access other DBMS (DB2 or
non-DB2) or do you mean DB2 ESE with DPF (shared nothing paralelism with
a single-system image, i.e. one DB2 database with partitioned tables)?

Cheers
Serge


No offense, Serge, but shared nothing parallelism means 1 partition per
physical node. No shared memory, no shared processors, no shared disk
controllers, no shared network connections, etc between partitions. I am not
necessarily suggesting that shared nothing is the optimal configuration when
considering price vs performance. At one time (more than 5 years ago) SMP,
and disk subsystem were much less sophisticated than today, and shared
nothing had clear benefits. Those days are gone.
Nov 12 '05 #3

P: n/a
Serge:

DB2 ESE with DPF (shared nothing paralelism with
a single-system image, i.e. one DB2 database with partitioned tables)

Basically a clustered environment.

--
BadPony
'03 ZRX1200R
'78 XS 1100
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
Bad Pony,

Do you mean federated as in using Nicknames to access other DBMS (DB2 or
non-DB2) or do you mean DB2 ESE with DPF (shared nothing paralelism with
a single-system image, i.e. one DB2 database with partitioned tables)?

Cheers
Serge

Nov 12 '05 #4

P: n/a
Mark,

None taken. I don't see what set you of though. The poster seemed to
equate federated with partitioned which is not the same.
The whole purpose of my post was asking for clarification.
Does the poster want to run Peoplesoft on DB2 II or on DB2 ESE + DPF?
Anyway we know now (s)he talks about DPF :-)

Cheers
Serge
Nov 12 '05 #5

P: n/a
OK, why would you want to run Peoplesoft in a clustered environment?
(other than to make an IBM salesperson rich and help pay my salary)
Is this a capacity/size or a failover issue?

Cheers
Serge
Nov 12 '05 #6

P: n/a
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
Mark,

None taken. I don't see what set you of though. The poster seemed to
equate federated with partitioned which is not the same.
The whole purpose of my post was asking for clarification.
Does the poster want to run Peoplesoft on DB2 II or on DB2 ESE + DPF?
Anyway we know now (s)he talks about DPF :-)

Cheers
Serge


I was only trying to point out that partitioned parallel databases (such as
implemented via DB2 ESE + DPF) are not necessarily "share nothing." The
concept of share nothing (one partition for each physical node first
implemented by Teradata many years ago) is one possible hardware
configuration of a partitioned database. Since the vast improvement of SMP,
shared disk sub-systems (even if the physical disks themselves are not
shared), true share nothing nodes are not used very often anymore, although
it is certainly possible to do it.

There seems to be a lot of confusion about "share nothing."
Nov 12 '05 #7

P: n/a
Hi Serge

Just looking for options, that's all. Have you ever done this, or know
anyone that has? If yes, I'd love to hear about any warstories you may have,
if not, but know someone who has, then please let me know.
Thanks

--
BadPony
'03 ZRX1200R
'78 XS 1100
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
OK, why would you want to run Peoplesoft in a clustered environment?
(other than to make an IBM salesperson rich and help pay my salary)
Is this a capacity/size or a failover issue?

Cheers
Serge

Nov 12 '05 #8

P: n/a
"BadPony" <eb**@ix.netcom.com> wrote in message
news:Jp********************@comcast.com...
Hi Serge

Just looking for options, that's all. Have you ever done this, or know
anyone that has? If yes, I'd love to hear about any warstories you may have, if not, but know someone who has, then please let me know.
Thanks

--
BadPony
'03 ZRX1200R
'78 XS 1100


What do you think you might gain by running Peoplesoft in a parallel
environment?
Nov 12 '05 #9

P: n/a
A question of definition I suppose. I suppose for the reasons you stated
commonly "shared nothing" refers (or at least I refer) to logical
ownership. I.e a node will neither share memory, nor data with any other
node.
You are correct that today this does neither mean that the nodes may not
share the same box right down to real RAM a single CPU and the hard drive.
I.e. I can run DB2 with several nodes (shared nothing) on my Laptop and
I do that on occasion.

Interestingly thsi means that the lines beween "shared disk" and "shared
nothing" contiously blur. E.g. while in Oracle RAC ("shared disk") each
node can get any data, there is AFAIK still one node that "controls" a
given pice of data at least from a locking perspective. (resulting in a
redistribution of the control when a node goes down).
(I'm sure if I'm wrong Daniel M. and Mark T. will not hesitate to
correct me :-)
Similarly in DB2 a failing node can be replaced by another node.
The truth (or let's call it winning architecture), in the end, will be
somewhere in the middle I'd guess and it will be neither shared disk,
nor shared nothing.

Cheers
Serge
Nov 12 '05 #10

P: n/a
Bad Pony,

I haven't but I'm a DB2 developer, not a consultant.
Note that IBM is pushing DPF for BI which is not what Peoplesoft does AFAIK.
An SAP BW would be a different story alltogether.

Cheers
Serge
Nov 12 '05 #11

P: n/a
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
A question of definition I suppose. I suppose for the reasons you stated
commonly "shared nothing" refers (or at least I refer) to logical
ownership. I.e a node will neither share memory, nor data with any other
node.
You are correct that today this does neither mean that the nodes may not
share the same box right down to real RAM a single CPU and the hard drive. I.e. I can run DB2 with several nodes (shared nothing) on my Laptop and
I do that on occasion.

Interestingly thsi means that the lines beween "shared disk" and "shared
nothing" contiously blur. E.g. while in Oracle RAC ("shared disk") each
node can get any data, there is AFAIK still one node that "controls" a
given pice of data at least from a locking perspective. (resulting in a
redistribution of the control when a node goes down).
(I'm sure if I'm wrong Daniel M. and Mark T. will not hesitate to
correct me :-)
Similarly in DB2 a failing node can be replaced by another node.
The truth (or let's call it winning architecture), in the end, will be
somewhere in the middle I'd guess and it will be neither shared disk,
nor shared nothing.

Cheers
Serge


Share nothing parallel architecture has no real meaning in for a single
partition database on a single node. That makes no sense because there are
not multiple partitions that share hardware resources. "Share nothing" or
"share whatever" only makes sense in the context of a multi-partition
database.

I didn't mean to suggest that in a share nothing environment, that
nodes/partitions don't sometimes share data, such as when a failover happens
or when join takes place that causes data to be transferred from one
node/partition to another. That really has nothing to do with whether an
architecture is share nothing, since the partitioning scheme of the data is
determined by DBA and not by the hardware or database software.

Early implementations of share nothing parallel database where limited by
the hardware. The last "forced" share nothing architecture from Teradata was
implemented on nodes that consisted of Intel 486 PC's running a propriety
16-bit OS that was incapable of sharing hardware resources. The first
implementations of DB2 PE (Parallel Edition) were implemented on single
processor RS/6000 nodes (AIX SMP was not available at that time IIRC), so
one DB2 partition per physical node was the norm.

The point is that share nothing is the relationship between the hardware
node and the logical partition. Most large scale IW parallel databases have
moved away from shared nothing because for price/performance reasons, as the
improvements in hardware and SMP operating systems makes at least some
sharing much more efficient and scalable. Although it is certainly true in
terms of absolute performance, that shared nothing does scale in a more
linear way when extremely large databases are needed. But shared nothing is
usually not the best in terms of price/performance.

Interestingly, recent hardware solutions provided with the IBM eServer 325
(and perhaps other vendors) has rekindled interest in share nothing (or
share less) implementations because of the affordability of AMD 64 bit nodes
on IBM blades. The IBM benchmarks for TPC-H used 2 partitions per physical
node (with a dual processor), but it would have been interesting to see the
difference in absolute performance and price/performance if they had only
used on partition per physical node.

Note that if one only creates a single partition on a dual-processor (or
more) physical node (blade), that is not sharing. The CPU's are not shared
with any other partition.
Nov 12 '05 #12

P: n/a
> E.g. while in Oracle RAC ("shared disk") each
node can get any data, there is AFAIK still one node that "controls" a
given pice of data at least from a locking perspective. (resulting in a
redistribution of the control when a node goes down).
Hmm - a little understated I think. A data block in a nodes memory will
have the associated locking information in the same memory as well. The
same data block can also be in memory on different nodes - each with
it's own lock information.

It's only when the the data is being updated that a special write lock
is migrated to or "mastered on" the node that is actually doing the
update. This is the Cache Fusion aspect of RAC (basically we use shared
memory techniques to do this).

However, there is not a single mastering node, as all nodes can be
masters of different bits of data at the same time.

If a node goes down, then any data blocks in that node's memory that
were dirty from writes need to be rebuilt and remastered on the
surviving nodes (maybe - if nobody wants them, you can do it later). So
hot blocks are read from disk (or, given that they are hot, will
probably already be in the recovering nodes memory), redo is applied,
and on you go.

Interestingly enough, the more nodes, the less work per node that needs
to be done if a node goes down. Makes the low cost blade environments
Mark alluded to a very, very interesting proposition.
(I'm sure if I'm wrong Daniel M. and Mark T. will not hesitate to
correct me :-)
Similarly in DB2 a failing node can be replaced by another node.


Well, there is actually no similarity between the time Oracle RAC takes
to re-build the dirty blocks from a lost node, and the time required for
a node to takeover another node's disk. The outage on one is measured in
seconds, the other isn't.

Nov 12 '05 #13

P: n/a
Mark A wrote:
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
A question of definition I suppose. I suppose for the reasons you stated
commonly "shared nothing" refers (or at least I refer) to logical
ownership. I.e a node will neither share memory, nor data with any other
node.
You are correct that today this does neither mean that the nodes may not
share the same box right down to real RAM a single CPU and the hard


drive.
I.e. I can run DB2 with several nodes (shared nothing) on my Laptop and
I do that on occasion.

Interestingly thsi means that the lines beween "shared disk" and "shared
nothing" contiously blur. E.g. while in Oracle RAC ("shared disk") each
node can get any data, there is AFAIK still one node that "controls" a
given pice of data at least from a locking perspective. (resulting in a
redistribution of the control when a node goes down).
(I'm sure if I'm wrong Daniel M. and Mark T. will not hesitate to
correct me :-)
Similarly in DB2 a failing node can be replaced by another node.
The truth (or let's call it winning architecture), in the end, will be
somewhere in the middle I'd guess and it will be neither shared disk,
nor shared nothing.

Cheers
Serge

Share nothing parallel architecture has no real meaning in for a single
partition database on a single node. That makes no sense because there are
not multiple partitions that share hardware resources. "Share nothing" or
"share whatever" only makes sense in the context of a multi-partition
database.

I didn't mean to suggest that in a share nothing environment, that
nodes/partitions don't sometimes share data, such as when a failover happens
or when join takes place that causes data to be transferred from one
node/partition to another. That really has nothing to do with whether an
architecture is share nothing, since the partitioning scheme of the data is
determined by DBA and not by the hardware or database software.

Early implementations of share nothing parallel database where limited by
the hardware. The last "forced" share nothing architecture from Teradata was
implemented on nodes that consisted of Intel 486 PC's running a propriety
16-bit OS that was incapable of sharing hardware resources. The first
implementations of DB2 PE (Parallel Edition) were implemented on single
processor RS/6000 nodes (AIX SMP was not available at that time IIRC), so
one DB2 partition per physical node was the norm.

The point is that share nothing is the relationship between the hardware
node and the logical partition. Most large scale IW parallel databases have
moved away from shared nothing because for price/performance reasons, as the
improvements in hardware and SMP operating systems makes at least some
sharing much more efficient and scalable. Although it is certainly true in
terms of absolute performance, that shared nothing does scale in a more
linear way when extremely large databases are needed. But shared nothing is
usually not the best in terms of price/performance.

Interestingly, recent hardware solutions provided with the IBM eServer 325
(and perhaps other vendors) has rekindled interest in share nothing (or
share less) implementations because of the affordability of AMD 64 bit nodes
on IBM blades. The IBM benchmarks for TPC-H used 2 partitions per physical
node (with a dual processor), but it would have been interesting to see the
difference in absolute performance and price/performance if they had only
used on partition per physical node.

Note that if one only creates a single partition on a dual-processor (or
more) physical node (blade), that is not sharing. The CPU's are not shared
with any other partition.


The only two commecial RDBMS products with shared nothing architecture
are DB2 on mainframes (not other platforms) and Oracle on all platforms.
It will be nice when IBM can implement this for DB2 on all platforms.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #14

P: n/a
Mark Townsend wrote:
> E.g. while in Oracle RAC ("shared disk") each

node can get any data, there is AFAIK still one node that "controls" a
given pice of data at least from a locking perspective. (resulting in
a redistribution of the control when a node goes down).

Hmm - a little understated I think. A data block in a nodes memory will
have the associated locking information in the same memory as well. The
same data block can also be in memory on different nodes - each with
it's own lock information.

It's only when the the data is being updated that a special write lock
is migrated to or "mastered on" the node that is actually doing the
update. This is the Cache Fusion aspect of RAC (basically we use shared
memory techniques to do this).

However, there is not a single mastering node, as all nodes can be
masters of different bits of data at the same time.

If a node goes down, then any data blocks in that node's memory that
were dirty from writes need to be rebuilt and remastered on the
surviving nodes (maybe - if nobody wants them, you can do it later). So
hot blocks are read from disk (or, given that they are hot, will
probably already be in the recovering nodes memory), redo is applied,
and on you go.

Interestingly enough, the more nodes, the less work per node that needs
to be done if a node goes down. Makes the low cost blade environments
Mark alluded to a very, very interesting proposition.
(I'm sure if I'm wrong Daniel M. and Mark T. will not hesitate to
correct me :-)
Similarly in DB2 a failing node can be replaced by another node.

Well, there is actually no similarity between the time Oracle RAC takes
to re-build the dirty blocks from a lost node, and the time required for
a node to takeover another node's disk. The outage on one is measured in
seconds, the other isn't.


I'm flattered that someone mentioned us by name "Daniel M. and Mark T.
will not hesitate" and we didn't.

Taught a RAC class just today ... we created the equivalent of a 6 CPU
mid-size box with a few thousand dollars worth of commodity Intel-based
hardware, a couple switches, and a NetApp NAS filer head. Not just
scalability but fail-over. One at a time we pulled the plug on 5 of the
machines and each time, with only a 15-20 seconds delay, the load was
redistributed until number six took over the entire load and not a
single transaction rolled back. Ah beauty of shared nothing.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #15

P: n/a
I'll let that last one go or we'll have to nest OTs....
Actually I let that whole thread go.

Cheers
Serge

PS: Mark: if you folks don't deliver O10g within a week you'll have an
expired TPC-C result on your hands.. What's up?
Nov 12 '05 #16

P: n/a
I trust you just confused a few words here.... DB2 z/Series is shared
disk and so is Oracle RAC.
Nov 12 '05 #17

P: n/a
Serge Rielau wrote:
I'll let that last one go or we'll have to nest OTs....
Actually I let that whole thread go.

Cheers
Serge

PS: Mark: if you folks don't deliver O10g within a week you'll have an
expired TPC-C result on your hands.. What's up?


I've got 10g ... don't you? ;-)

In a few weeks I'll send you my copy.

Good question Serge ... yeah Mark ... What's up? ;-)

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #18

P: n/a
Serge Rielau wrote:
I trust you just confused a few words here.... DB2 z/Series is shared
disk and so is Oracle RAC.


Old syntax habits die hard.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #19

P: n/a
"Daniel Morgan" <da******@x.washington.edu> wrote in message
news:1075003854.992919@yasure...

The only two commecial RDBMS products with shared nothing architecture
are DB2 on mainframes (not other platforms) and Oracle on all platforms.
It will be nice when IBM can implement this for DB2 on all platforms.

--
Daniel Morgan


DB2 for Linux, UNIX and Windows clearly supports shared nothing
architecture. This is really a hardware issue, not a software issue. AFAIK,
all multi-node parallel databases that support multiple nodes, also support
shared nothing hardware (one database partition per physical node), although
it is often not cost effective to implement that way.
Nov 12 '05 #20

P: n/a
Know anyone that has done it using SAP?

--
BadPony
'03 ZRX1200R
'78 XS 1100
"Serge Rielau" <sr*****@ca.eye-be-em.com> wrote in message
news:bu**********@hanover.torolab.ibm.com...
Bad Pony,

I haven't but I'm a DB2 developer, not a consultant.
Note that IBM is pushing DPF for BI which is not what Peoplesoft does AFAIK. An SAP BW would be a different story alltogether.

Cheers
Serge

Nov 12 '05 #21

P: n/a
Mark A wrote:
"Daniel Morgan" <da******@x.washington.edu> wrote in message
news:1075003854.992919@yasure...
The only two commecial RDBMS products with shared nothing architecture
are DB2 on mainframes (not other platforms) and Oracle on all platforms.
It will be nice when IBM can implement this for DB2 on all platforms.

--
Daniel Morgan

DB2 for Linux, UNIX and Windows clearly supports shared nothing
architecture. This is really a hardware issue, not a software issue. AFAIK,
all multi-node parallel databases that support multiple nodes, also support
shared nothing hardware (one database partition per physical node), although
it is often not cost effective to implement that way.


The difference is that with Oracle there is only one database no
matter how many nodes. Oracle does not replicate (like SQL Server) or
partition its data into separate databases (like DB2).

The equality you imply with "one database partition per physical node"
is where the architecture difference is important. Oracle never creates
one database partition per physical node unless you don't understand
what you are trying to do.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #22

P: n/a
Daniel,

Mark A. and I may differ on the intricate details, but your definition
for shared nothing is way out there.
Because Oracle does NOT partition the data it certainly NOT shared
nothing and Oracle's docs go through great lengths to beg to differ from
shared nothing.
Note that I'm not saying anything about what's good or bad. I'm just
saying that Oracle in NOT doing shared nothing. Oracle's approach is
commonly referred to as shared disk since every node has access to all
the data (just as you say).

You can classify loosely like this:

shared nothing (one DB across multiple nodes, nodes own data):
DB2 ESE + DPF,
Informix XPS
Teradata

shared disk (one DB across multiple nodes, nodes do not own data):
Oracle with RAC
DB2 z/Series

federated (multiple DBs linked together presenting a single DB view)
MS SQL Server with linked tables (did i get that right?) (used for
distributed partitioned views)
DB2 II (Nicknames)
Informix IDS (external tables (?))
Oracle + Transparent gateway

Somewhere between federated and shared nothing DB2 for iSeries fits in
as well.
Nov 12 '05 #23

P: n/a
Serge Rielau wrote:
Daniel,

Mark A. and I may differ on the intricate details, but your definition
for shared nothing is way out there.


Not just "way out there" ... dead wrong.

It was late at night and I'd just spent the previous 12 hours teaching a
RAC class. Oracle hasn't used Shared Nothing for years and the last time
was, IIRC, with OPS. Shared Nothing has nothing to do with RAC.

My apology for having too little sleep and too much Chimay.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #24

P: n/a
PS: Mark: if you folks don't deliver O10g within a week you'll have an
expired TPC-C result on your hands.. What's up?
I guess "we" tend to not subscribe to the the
there is significant pressure to get function out.
and
It's a well known problem that testing can never we enough.


school of software development.

Nov 12 '05 #25

P: n/a
Daniel Morgan <da******@x.washington.edu> wrote in message news:<1075044526.452560@yasure>...
The difference is that with Oracle there is only one database no
matter how many nodes. Oracle does not replicate (like SQL Server) or
partition its data into separate databases (like DB2).


The data is partitioned, but the partitions are still part of the same
database. I believe Teradata uses a similar approach. Shared-nothing
clustered architectures still dominate the TPC-H performance rankings
(which are noticeably RAC-free).
DG
Nov 12 '05 #26

P: n/a
Mark Townsend wrote, with liberal editing:

....
I guess "we" tend to not subscribe to the the
there is significant pressure to get function out.


Yes, you'll find that there is less pressure to get function out as your
market share falls.

Nov 12 '05 #27

P: n/a
> > The difference is that with Oracle there is only one database no
matter how many nodes. Oracle does not replicate (like SQL Server) or
partition its data into separate databases (like DB2).


The data is partitioned, but the partitions are still part of the same
database. I believe Teradata uses a similar approach. Shared-nothing
clustered architectures still dominate the TPC-H performance rankings
(which are noticeably RAC-free).

DG


Teradata invented share nothing (and also parallel database partitioning).
But it is rarely used today because it is usually more effective to have
multiple partitions per physical node, even when there are multiple nodes.
Back when Teradata invented parallel database partitioning, the hardware and
software platforms were not capable of SMP, so share nothing was the only
option. Today's sophisticated OS's and hardware can use SMP quite
efficiently.

Share nothing usually does not yield the best price/performance, but it
scales in linear fashion for each additional partition. Today, most parallel
implementations use a combination of multiple physical nodes, with multiple
partitions per node (SMP).
Nov 12 '05 #28

P: n/a
Database Guy wrote:
Daniel Morgan <da******@x.washington.edu> wrote in message news:<1075044526.452560@yasure>...

The difference is that with Oracle there is only one database no
matter how many nodes. Oracle does not replicate (like SQL Server) or
partition its data into separate databases (like DB2).

The data is partitioned, but the partitions are still part of the same
database. I believe Teradata uses a similar approach. Shared-nothing
clustered architectures still dominate the TPC-H performance rankings
(which are noticeably RAC-free).
DG


Add a node with shared nothing and your mean time to failure goes down
.... not up. Add a node with shared nothing and your realized performance
per node goes down rather than remaining stable. Add or subtract a node
with shared everything ... and you don't need to rewrite a single line
of code, perform a shutdown, reparse and reload your data.

Last time I looked TPC didn't give a .... about maintainability.

The fact that they are part of the same database is meaningless. If you
lose the disks attached to node 6 of 10 what happens? Is that data live
anywhere else? Do the transactions seamlessly fail-over and continue
running?

And , IIRC, DB2 on mainframes is shared everything ... not shared nothing.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #29

P: n/a
Blair Adamache wrote:
Mark Townsend wrote, with liberal editing:

...
I guess "we" tend to not subscribe to the the
there is significant pressure to get function out.



Yes, you'll find that there is less pressure to get function out as your
market share falls.


Actually the opposite is true. A fact that seems to be recognied by
those folks on Wall Street where it appears that Larry's stock closed
above its previous 52 week high on Friday. Hardly a sign of corporate
failure.

But in this case ... Oracle's statement about the release of 10g by the
end of the year was related to Oracle's fiscal year which ends in May
.... not your calendar year. From what I hear ... it will be on-time.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #30

P: n/a
"Daniel Morgan" <da******@x.washington.edu> wrote in message
news:1075072958.276429@yasure...
Add a node with shared nothing and your mean time to failure goes down
... not up. Add a node with shared nothing and your realized performance
per node goes down rather than remaining stable. Add or subtract a node
with shared everything ... and you don't need to rewrite a single line
of code, perform a shutdown, reparse and reload your data.
Shared nothing as implemented for parallel databases was specifically
created for very large IW databases where uptime was not as critical as
performance for large complex queries that typically access the entire
table. However, the use of RAID disk technology makes complete outages
somewhat unlikely.
Last time I looked TPC didn't give a .... about maintainability.
TPC is the Transaction Processing Performance Council, so maintainability is
not the primary consideration, especially since it cannot be objectively
measured. No one would argue that TPC benchmarks should be the sole criteria
for choosing a database product. For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.

TPC benchmarks do, however, require fairly stringent rules about maintaining
data integrity (logging, transaction recovery, roll forward recovery, etc)
and concurrency (multiple users hitting the database at the same time) that
would be typical for a real-world production database.

The fact that they are part of the same database is meaningless. If you
lose the disks attached to node 6 of 10 what happens? Is that data live
anywhere else? Do the transactions seamlessly fail-over and continue
running?
Shared nothing was not designed for transactions. It was designed for
massive IW databases.
And , IIRC, DB2 on mainframes is shared everything ... not shared nothing.

That is basically correct.
Nov 12 '05 #31

P: n/a
> For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.


Prove it. And don't quote me a lot of fly by night supposed analysts
whose main income seems to come from talking at DB2 user conferences.

Nov 12 '05 #32

P: n/a
Blair Adamache wrote:
Mark Townsend wrote, with liberal editing:

...
I guess "we" tend to not subscribe to the the
there is significant pressure to get function out.



Yes, you'll find that there is less pressure to get function out as your
market share falls.

Great, so you'll soon have some time to fix some bugs then.

Nov 12 '05 #33

P: n/a
Hi

But still there is one vital difference the thread failed to address..

shared nothing --Biggest advantage is it is transparent to
applications...You never know from which node you get your data.

shared everything--You need to partition your application before
hand..as when you query

1)select * from dept where deptno=10 it wil go to node 1

2)select * from dept where deptno=20 it will go to node 2

well all those database that use shared nothing architecture (The
programmers have worked very very hard there ) have made the lives of
application developers much easier..

with shared everything..the application will have to be partitoned to
take advantage of the underlying shared everything architecture..

regards
Hrishy
"Mark A" <ma@switchboard.net> wrote in message news:<1s***************@news.uswest.net>...
"Daniel Morgan" <da******@x.washington.edu> wrote in message
news:1075072958.276429@yasure...
Add a node with shared nothing and your mean time to failure goes down
... not up. Add a node with shared nothing and your realized performance
per node goes down rather than remaining stable. Add or subtract a node
with shared everything ... and you don't need to rewrite a single line
of code, perform a shutdown, reparse and reload your data.

Shared nothing as implemented for parallel databases was specifically
created for very large IW databases where uptime was not as critical as
performance for large complex queries that typically access the entire
table. However, the use of RAID disk technology makes complete outages
somewhat unlikely.

Last time I looked TPC didn't give a .... about maintainability.

TPC is the Transaction Processing Performance Council, so maintainability is
not the primary consideration, especially since it cannot be objectively
measured. No one would argue that TPC benchmarks should be the sole criteria
for choosing a database product. For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.

TPC benchmarks do, however, require fairly stringent rules about maintaining
data integrity (logging, transaction recovery, roll forward recovery, etc)
and concurrency (multiple users hitting the database at the same time) that
would be typical for a real-world production database.

The fact that they are part of the same database is meaningless. If you
lose the disks attached to node 6 of 10 what happens? Is that data live
anywhere else? Do the transactions seamlessly fail-over and continue
running?

Shared nothing was not designed for transactions. It was designed for
massive IW databases.
And , IIRC, DB2 on mainframes is shared everything ... not shared nothing.

That is basically correct.

Nov 12 '05 #34

P: n/a
> > For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.


"Mark Townsend" <ma***********@comcast.net> wrote in message
news:40**************@comcast.net... Prove it. And don't quote me a lot of fly by night supposed analysts
whose main income seems to come from talking at DB2 user conferences.

I don't think there is any way to "prove" it. My comments where based on my
personal experience, not what anyone else has told me.

I once hired an Oracle Consultant (form Oracle Corporation) to do a simple
single node basic install (no remote clients of any kind) and it took 2 days
(at $1400 per day). On another occasion, it took some other DBA's I hired a
week to get Oracle installed. About 5 years ago, I tried to install Oracle
on Windows and gave up after a few weeks (working part time on the install).

DB2 can normally be installed in an hour or two by a chimpanzee.

I will have to admit, that Oracle is a much more complex and feature rich
product, but I also think that cost of ownership is very high, especially
when it comes to personnel.
Nov 12 '05 #35

P: n/a
Mark Townsend wrote:
> For example, Oracle requires more DBA

support than DB2, an important consideration when choosing.

Prove it. And don't quote me a lot of fly by night supposed analysts
whose main income seems to come from talking at DB2 user conferences.


Apparently another one of those that still thinks we run Oracle
databases from the command line. Perhaps we can provide them with
some sort of sedative if they ever see ASM and ADDM and become
agitated.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #36

P: n/a
Comments in-line.

hrishy wrote:
Hi

But still there is one vital difference the thread failed to address..

shared nothing --Biggest advantage is it is transparent to
applications...You never know from which node you get your data.
Which amazingly enough is the same as shared everything except with
shared everything it is not only transparent ... it is meaningless
because the data is not associated with a node.
shared everything--You need to partition your application before
hand..as when you query
What have you been smoking? Today I taught a RAC class. Set up three
two-node clusters from scratch. We had 6 PCs, two switches, and one F810
NetApp NAS filer head. All data in a single datafile (meaning a single
tablespace) and accessed it seemlessly from all nodes of all three
clusters. Then sequentially pulled the plug on half the nodes and
watched the transactions fail-over in less than a minute.

Did we partition any data? Not one byte. I'd suggest you learn the
available architectures, federated, shared nothing, and shared
everything before trying to describe how they work.
1)select * from dept where deptno=10 it wil go to node 1
Not with shared everything.
2)select * from dept where deptno=20 it will go to node 2
Not with shared everything.

What goes to a node is based on load balancing and nothing else. Too
much load on node 1 ... transaction runs on 2.
well all those database that use shared nothing architecture (The
programmers have worked very very hard there ) have made the lives of
application developers much easier..
Right up until they add or subtract a node. With shared everything the
developers aren't involved in writing a single line of code beyon
stitching in TAF.
with shared everything..the application will have to be partitoned to
take advantage of the underlying shared everything architecture..

regards
Hrishy


What you don't understand would fill a book. Luckily that book ... has
already been written by Mike Ault.

http://www.amazon.com/exec/obidos/tg...glance&s=books

It is the one I use to teach classes at the University of Washington.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #37

P: n/a
>
DB2 can normally be installed in an hour or two by a chimpanzee.


I guess we don't get a lot of chimpanzee's installing the software, but
it's not difficult at all and can indeed be done by a reasonably clever
9 year old - see the first in the thread at http://tinyurl.com/32gs7

So tell me, how does Global Services actually get the chimps to wear the
suits ?

Nov 12 '05 #38

P: n/a
> > DB2 can normally be installed in an hour or two by a chimpanzee.


I guess we don't get a lot of chimpanzee's installing the software, but
it's not difficult at all and can indeed be done by a reasonably clever
9 year old - see the first in the thread at http://tinyurl.com/32gs7

So tell me, how does Global Services actually get the chimps to wear the
suits ?

I am self employed, so I can't answer that one. I just want to know how
Oracle Consulting can charge $1400 per day to do a very simple Oracle
install and why does it take 2 full days.
Nov 12 '05 #39

P: n/a
Mark A wrote:
For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.


"Mark Townsend" <ma***********@comcast.net> wrote in message
news:40**************@comcast.net...
Prove it. And don't quote me a lot of fly by night supposed analysts
whose main income seems to come from talking at DB2 user conferences.


I don't think there is any way to "prove" it. My comments where based on my
personal experience, not what anyone else has told me.

I once hired an Oracle Consultant (form Oracle Corporation) to do a simple
single node basic install (no remote clients of any kind) and it took 2 days
(at $1400 per day). On another occasion, it took some other DBA's I hired a
week to get Oracle installed. About 5 years ago, I tried to install Oracle
on Windows and gave up after a few weeks (working part time on the install).

DB2 can normally be installed in an hour or two by a chimpanzee.

I will have to admit, that Oracle is a much more complex and feature rich
product, but I also think that cost of ownership is very high, especially
when it comes to personnel.


Sorry Mark but academic freedom demands that I be honest here.

Anyone that took two days to install a single node shouldn't have made
it through the interview process unless your hardware and operating
systems were junk.

Today I did the following between 9am and 5:30pm:

Wired a network with two switches, a router, and three separate
networks. Installed Linux EL AS onto six machines from CD, installed the
Oracle installer and clusterware 9.2.0.1 and patched it to 9.2.0.4 on
the same six machines. Installed the Oracle software at 9.2.0.1 on all
six machines and patched it to 9.2.0.4 and then created three separate
databases including all tablespaces and tested fail-over from three RAC
clusters using a TAF aware Java client. This was done by two instructors
and six students ... and it included more than two hours of
whiteboarding during the course.

So am I going to believe any competent person did what you claim? No.
The operative word here being 'competent'.

There are plenty of people working with any technically advanced
product, Oracle, DB2, Informix, Sybase, etc. that are marginal. Next
time hire better. I've got students that I'd fail if it took them two
days to install a node and they'd be really happy with the money you
paid ... almost their entire tuition for my class for just one day of work.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #40

P: n/a

"Daniel Morgan" <da******@x.washington.edu> wrote in message
news:1075099486.168786@yasure...
Mark A wrote:
For example, Oracle requires more DBA
support than DB2, an important consideration when choosing.


"Mark Townsend" <ma***********@comcast.net> wrote in message
news:40**************@comcast.net...
Prove it. And don't quote me a lot of fly by night supposed analysts
whose main income seems to come from talking at DB2 user conferences.


I don't think there is any way to "prove" it. My comments where based on my personal experience, not what anyone else has told me.

I once hired an Oracle Consultant (form Oracle Corporation) to do a simple single node basic install (no remote clients of any kind) and it took 2 days (at $1400 per day). On another occasion, it took some other DBA's I hired a week to get Oracle installed. About 5 years ago, I tried to install Oracle on Windows and gave up after a few weeks (working part time on the install).
DB2 can normally be installed in an hour or two by a chimpanzee.

I will have to admit, that Oracle is a much more complex and feature rich product, but I also think that cost of ownership is very high, especially when it comes to personnel.


Sorry Mark but academic freedom demands that I be honest here.

Anyone that took two days to install a single node shouldn't have made
it through the interview process unless your hardware and operating
systems were junk.

Interview process? I hired a consultant from Oracle Corporation and paid top
dollar because I wanted a qualified and experienced person and I needed it
done quickly. There was nothing wrong with the hardware or software on that
system. I gave the Oracle consultant full time use of my UNIX system
administrator during the 2 days.
Nov 12 '05 #41

P: n/a
I just want to know how
Oracle Consulting can charge $1400 per day to do a very simple Oracle
install
I really don't know. Did they doc the environment ? Was there OS patches
that needed to be applied ? Did they set up best practices for backup
and recovery etc. Any of those could take a day or two. If you want to
give me a PO I'll look it up from this end.
and why does it take 2 full days.


The install doesn't. Even on my iBook where I am writing this, a full
install with manual creation of the database takes less than 1 hour. On
my Windows laptop with the seed database, around 20-30 minutes.

Nov 12 '05 #42

P: n/a
Mark A wrote:
DB2 can normally be installed in an hour or two by a chimpanzee.


I guess we don't get a lot of chimpanzee's installing the software, but
it's not difficult at all and can indeed be done by a reasonably clever
9 year old - see the first in the thread at http://tinyurl.com/32gs7

So tell me, how does Global Services actually get the chimps to wear the
suits ?


I am self employed, so I can't answer that one. I just want to know how
Oracle Consulting can charge $1400 per day to do a very simple Oracle
install and why does it take 2 full days.


Far be it from me to defend Oracle Consulting as I have come in after
them a few times myself: Enough said. But without a thorough knowledge
of what that person found when they walked in the door it is hard to say.

The first time I did a contract at Boeing I spent the first three weeks
waiting for them to get me a desk and a computer ... at my full
consulting rate. I may be ethical but I'm not a fool. I'd have to hear
the consultants description of the job as perhaps your lack of knowledge
about Oracle caused a simple job to be a trail of patience getting
simple things like network access and passwords.

But as I said far be it from me to defend Oracle Consulting. All
consulting companies have marginal employees: All of them. Perhaps you
got one. Can't say. Next time call me. I'll send one of my students. You
cover one quarter's tuition and the plane ticket and I know I'd have a
half-dozen volunteers.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #43

P: n/a
Mark A wrote:
Interview process? I hired a consultant from Oracle Corporation and paid top
dollar because I wanted a qualified and experienced person and I needed it
done quickly. There was nothing wrong with the hardware or software on that
system. I gave the Oracle consultant full time use of my UNIX system
administrator during the 2 days.


Assuming what you say is true ... I'd send Mark the invoice number he
requested because it is simply outrageous. I've personally installed
Oracle 9.2.0.1, patched it to 9.2.0.4, and built a database on my PIII
500MHz IBM 600X Thinkpad in less than two hours. Send the invoice number
to Mark Townsend and do what you should have done in the first place ...
not complain ... demand an explanation.

Once again assuming, there isn't more to the story, the problem is not
the Oracle software ... it is the consultant and should not be
tolerated. There is, however, one proviso. How big were the tablespaces
he was creating? It can take hours to lay down hundreds of gigabytes of
datafile. If that is the case the limiting factor is not Oracle ... but
rather your DASD.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #44

P: n/a
> > I just want to know how
Oracle Consulting can charge $1400 per day to do a very simple Oracle
install


I really don't know. Did they doc the environment ? Was there OS patches
that needed to be applied ? Did they set up best practices for backup
and recovery etc. Any of those could take a day or two. If you want to
give me a PO I'll look it up from this end.
and why does it take 2 full days.


The install doesn't. Even on my iBook where I am writing this, a full
install with manual creation of the database takes less than 1 hour. On
my Windows laptop with the seed database, around 20-30 minutes.
"Mark Townsend


This was a simple install for a demo of a software product that used
Oracle. It was not a production server and was not even permanent. The
reason it took 2 days is because the Oracle consultant had problems doing
the install, just like most Oracle installs I have seen. I am not trying to
berate Oracle other than to say that it is a much more complex product than
DB2 and usually requires more staff.

I don't have access to the PO. It was about 2 years ago.
Nov 12 '05 #45

P: n/a
> Mark A wrote:
Interview process? I hired a consultant from Oracle Corporation and paid top dollar because I wanted a qualified and experienced person and I needed it done quickly. There was nothing wrong with the hardware or software on that system. I gave the Oracle consultant full time use of my UNIX system
administrator during the 2 days.


Assuming what you say is true ... I'd send Mark the invoice number he
requested because it is simply outrageous. I've personally installed
Oracle 9.2.0.1, patched it to 9.2.0.4, and built a database on my PIII
500MHz IBM 600X Thinkpad in less than two hours. Send the invoice number
to Mark Townsend and do what you should have done in the first place ...
not complain ... demand an explanation.

Once again assuming, there isn't more to the story, the problem is not
the Oracle software ... it is the consultant and should not be
tolerated. There is, however, one proviso. How big were the tablespaces
he was creating? It can take hours to lay down hundreds of gigabytes of
datafile. If that is the case the limiting factor is not Oracle ... but
rather your DASD.

--
Daniel Morgan


As I have explained, the Oracle install was for a customer demo of a
software package. The database size was very small. In fact, I don't believe
that the install involved setting up any data, that was done by the
application install.

As I have said, my experience is that Oracle has a higher cost of ownership,
especially with regard to staff. YMMV.
Nov 12 '05 #46

P: n/a
Mark A wrote:
I just want to know how
Oracle Consulting can charge $1400 per day to do a very simple Oracle
install


I really don't know. Did they doc the environment ? Was there OS patches
that needed to be applied ? Did they set up best practices for backup
and recovery etc. Any of those could take a day or two. If you want to
give me a PO I'll look it up from this end.

and why does it take 2 full days.


The install doesn't. Even on my iBook where I am writing this, a full
install with manual creation of the database takes less than 1 hour. On
my Windows laptop with the seed database, around 20-30 minutes.
"Mark Townsend

This was a simple install for a demo of a software product that used
Oracle. It was not a production server and was not even permanent. The
reason it took 2 days is because the Oracle consultant had problems doing
the install, just like most Oracle installs I have seen. I am not trying to
berate Oracle other than to say that it is a much more complex product than
DB2 and usually requires more staff.

I don't have access to the PO. It was about 2 years ago.


Then (A) you should have objected immediately and (B) you should have
hired someone that knew what they were doing.

That someone comes from Oracle Consulting is no guarantee of expertise
anymore than the fact that someone comes from IBM consulting is a
guarantee. And at Boeing I worked with consultants from both companies,
sometimes on the same project, so .... well you can fill in the blank.

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #47

P: n/a
Mark A wrote:
Mark A wrote:

Interview process? I hired a consultant from Oracle Corporation and paid
top
dollar because I wanted a qualified and experienced person and I needed
it
done quickly. There was nothing wrong with the hardware or software on
that
system. I gave the Oracle consultant full time use of my UNIX system
administrator during the 2 days.


Assuming what you say is true ... I'd send Mark the invoice number he
requested because it is simply outrageous. I've personally installed
Oracle 9.2.0.1, patched it to 9.2.0.4, and built a database on my PIII
500MHz IBM 600X Thinkpad in less than two hours. Send the invoice number
to Mark Townsend and do what you should have done in the first place ...
not complain ... demand an explanation.

Once again assuming, there isn't more to the story, the problem is not
the Oracle software ... it is the consultant and should not be
tolerated. There is, however, one proviso. How big were the tablespaces
he was creating? It can take hours to lay down hundreds of gigabytes of
datafile. If that is the case the limiting factor is not Oracle ... but
rather your DASD.

--
Daniel Morgan

As I have explained, the Oracle install was for a customer demo of a
software package. The database size was very small. In fact, I don't believe
that the install involved setting up any data, that was done by the
application install.

As I have said, my experience is that Oracle has a higher cost of ownership,
especially with regard to staff. YMMV.


I'd say your "experience" is meaningless. Not from the standpoint that
it wasn't your experience. But rather the fact that the experience you
had bears no relationship to the experience you should have, or could
have, had. Any hardware not more than 4 years old, doing what you've
described, I'll do it in one hour max or do it for free. Of course I
still want the $1400/diem. ;-)

--
Daniel Morgan
http://www.outreach.washington.edu/e...ad/oad_crs.asp
http://www.outreach.washington.edu/e...oa/aoa_crs.asp
da******@x.washington.edu
(replace 'x' with a 'u' to reply)

Nov 12 '05 #48

P: n/a
Hi Daniel

I think i was misread..what you are referring to is shared
nothing...the RAC server which you setup is shared nothing
architecture..

I woudl like to sum up in case I was not clear..

shared nothing---(Hard for developers who wrote this architecture but
easy for application developers)

sahred everything ----(easy for developers who wrote this architecture
But difficult for applications developers as you now have to
prepartition data)

regards
Hrishy

Daniel Morgan <da******@x.washington.edu> wrote in message news:<1075097863.13589@yasure>...
Comments in-line.

hrishy wrote:
Hi

But still there is one vital difference the thread failed to address..

shared nothing --Biggest advantage is it is transparent to
applications...You never know from which node you get your data.


Which amazingly enough is the same as shared everything except with
shared everything it is not only transparent ... it is meaningless
because the data is not associated with a node.
shared everything--You need to partition your application before
hand..as when you query


What have you been smoking? Today I taught a RAC class. Set up three
two-node clusters from scratch. We had 6 PCs, two switches, and one F810
NetApp NAS filer head. All data in a single datafile (meaning a single
tablespace) and accessed it seemlessly from all nodes of all three
clusters. Then sequentially pulled the plug on half the nodes and
watched the transactions fail-over in less than a minute.

Did we partition any data? Not one byte. I'd suggest you learn the
available architectures, federated, shared nothing, and shared
everything before trying to describe how they work.
1)select * from dept where deptno=10 it wil go to node 1


Not with shared everything.
2)select * from dept where deptno=20 it will go to node 2


Not with shared everything.

What goes to a node is based on load balancing and nothing else. Too
much load on node 1 ... transaction runs on 2.
well all those database that use shared nothing architecture (The
programmers have worked very very hard there ) have made the lives of
application developers much easier..


Right up until they add or subtract a node. With shared everything the
developers aren't involved in writing a single line of code beyon
stitching in TAF.
with shared everything..the application will have to be partitoned to
take advantage of the underlying shared everything architecture..

regards
Hrishy


What you don't understand would fill a book. Luckily that book ... has
already been written by Mike Ault.

http://www.amazon.com/exec/obidos/tg...glance&s=books

It is the one I use to teach classes at the University of Washington.

Nov 12 '05 #49

P: n/a
Hi daniel
Opps sorry..i made a blunder while posting...:-)

My line shared everything should hhave read as shared nothing and
shared nothing should have read as sahred everything.. :-)

Thanks for correcting me.

so my lines should read

select * from dept where deptno=10 it wil go to node 1

with shared nothing
select * from dept where deptno=10 it wil go to any of the nodes

with shared everything..
so application partitioning becomes vital in shared nothing
architecture.

once again thanks for correcting me.

regards
Hrishy

Daniel Morgan <da******@x.washington.edu> wrote in message news:<1075097863.13589@yasure>...
Comments in-line.

hrishy wrote:
Hi

But still there is one vital difference the thread failed to address..

shared nothing --Biggest advantage is it is transparent to
applications...You never know from which node you get your data.


Which amazingly enough is the same as shared everything except with
shared everything it is not only transparent ... it is meaningless
because the data is not associated with a node.
shared everything--You need to partition your application before
hand..as when you query


What have you been smoking? Today I taught a RAC class. Set up three
two-node clusters from scratch. We had 6 PCs, two switches, and one F810
NetApp NAS filer head. All data in a single datafile (meaning a single
tablespace) and accessed it seemlessly from all nodes of all three
clusters. Then sequentially pulled the plug on half the nodes and
watched the transactions fail-over in less than a minute.

Did we partition any data? Not one byte. I'd suggest you learn the
available architectures, federated, shared nothing, and shared
everything before trying to describe how they work.
1)select * from dept where deptno=10 it wil go to node 1


Not with shared everything.
2)select * from dept where deptno=20 it will go to node 2


Not with shared everything.

What goes to a node is based on load balancing and nothing else. Too
much load on node 1 ... transaction runs on 2.
well all those database that use shared nothing architecture (The
programmers have worked very very hard there ) have made the lives of
application developers much easier..


Right up until they add or subtract a node. With shared everything the
developers aren't involved in writing a single line of code beyon
stitching in TAF.
with shared everything..the application will have to be partitoned to
take advantage of the underlying shared everything architecture..

regards
Hrishy


What you don't understand would fill a book. Luckily that book ... has
already been written by Mike Ault.

http://www.amazon.com/exec/obidos/tg...glance&s=books

It is the one I use to teach classes at the University of Washington.

Nov 12 '05 #50

96 Replies

This discussion thread is closed

Replies have been disabled for this discussion.